I had no clue it was this bad

Great file-manager, would recommend! Has a lot of plugins, integrates well with fzf to find files and zoxide for directories.

I use it as my default file manager and also in nvim.

Thanks for taking the time to share

I don’t think so. From what I gathered, the only thing Play Services can see on GrapheneOS is the list of other apps you have installed. That’s it. They can’t see anything else unless you grant access to it. You’re not giving Google root access to your phone, you’re just installing an app that happens to be made by Google, and it’s locked down like everything else.

Edit: https://youtu.be/YB01HHFitFA?t=625 I just saw this video apparently apps can still communicate with each other so you might want to isolate if that’s something you’re worried about.

Edit 2 : Another relevant link https://discuss.grapheneos.org/d/28558-google-can-still-see-my-app-activity-on-grapheneos/2

The threat level for google play services is different in graphene as it runs in what they call an “appbox,” which basically means Google Play is just another app that’s sandboxed like everything else.

One thing I haven’t understood properly I feel is how notifications work. They talked there’s basically 3 ways of sending notifications on android. FCM (googles system) , websockets, unifiedpush. Most apps use FCM so you need play services installed to get notifications, right?

How does that work through profiles though? Some commenter in this thread said you can forward them from another profile if that profile is running in the background? But if I have google play services installed on profile B but not profile A? Do I have to install them on every profile?

I may not fully understand how profiles work yet.

Yeah, as they said most banking apps now work, however, Google Pay doesn’t.

There are alternatives to it like curve pay but I haven’t done the research whether they’re trustworthy enough. EU company I think.

Yeah I apologize, I incorrectly assumed that GrapheneOS’s BFU state is more secure and requires you to enter your passphrase by default and not PIN and that this is not available on stock android which some people pointed out it is.

On a related note though, Graphene does have an interesting feature where if phone hasn’t been unlocked for some time it will force reboot to get into that BFU state. Metroplex sets it to 8 hours.

I think they also have some aggressive USB port control, but I haven’t looked into it. Where you can only charge phone in BFU state or something like that. Haven’t had time to read into it : https://grapheneos.org/features#usb-c-port-and-pogo-pins-control

Thanks for the in-depth answer, I think I will try installing Graphene today.

This can not only be turned off entirely in settings, but you can actually modify it on a per-network basis!

Oh nice ! Makes it way more useful then as I saw forum threads of people saying there’s no point in randomizing on your home network and may cause issues.

GrapheneOS’s airplane mode disables the cellular radio entirely, whereas some OEMs don’t do that on their phones, even when you turn on airplane mode, meaning your cell provider could still triangulate your position regardless of if you have airplane mode on or off.

Did not know that, fascinating! Even Airplane mode is upgraded :D

Yeah I noticed the main AUR package was last updated in June 2024. Thought they abandoned it but the GitHub shows the last release was around the same time. Downloaded sioyek-git instead and it works great.

I think I’m sticking with Sioyek. It checks enough boxes for what I need from a pdf viewer. Well documented, no performance issues, and it supports epub too.

The command line tools, portals, ruler for reading, keyboard text selection, searchable highlights, easy file opening, marking. Really vim-like. Need to customize some keybinds but otherwise don’t see a reason to look elsewhere for now.

Oh Sioyek looks interesting. Also the blog is great !

I personally want something more minimal.

It has a lot of momentum, so it will continue to dominate. But I wonder if it will decline over the long term as Linux continues to improve. Similar to how smartphones barely differentiate themselves from one another these days (compared to the past) maybe operating systems will have a similar fate. Maybe I’m a bit naive, but perhaps Linux will eventually have all the stability and ease of use of Windows, while also offering privacy, customization, and open-source benefits so there will be no real reason to use windows and the split will be more even.

Maybe… eventually…

I didn’t like it either on first play-through but I will try it again soon !

Not a safari user so don’t really know but on a quick google search did find this : https://github.com/televator-apps/vimari

On a related note, try Vimium (FF / chrome extension) that brings vim motions into your browser. You will have a more complete experience.

I don’t mess with code autocomplete, Cursor, agents or any of that stuff. I’ve got subscriptions to 2 platforms that give me access to a bunch of different models and I just ask whatever model I need directly, copy/paste the context it needs. On that note, AI search engines like Perplexity genuinely bring zero value to my workflow. I’d rather do the searching myself and feed it the relevant context, feels like it misleads me more often than it helps. I actually have a Perplexity sub (got it free) and haven’t touched their web search in like 4 months.

I’ve thought about the environmental impact and taken steps to minimize my usage. That’s actually one reason I avoid Cursor, agents, and AI web search - feels super wasteful and I’m not convinced it’s sustainable long-term. I guess I just like being in control, you know? I also try using smaller open source models when I can, even if they’re not as powerful.

My go-to models right now for daily use (easiest to hardest tasks): Llama 4 Scout -> DeepSeek v3.1 -> DeepSeek v3.1 (thinking) -> Gemini 2.5 Pro / Claude 4 Sonnet (thinking) -> GPT 5 (thinking). Sometimes I’ll throw in other models like Gemini 2.5 Flash but mostly stick to these.

By the way I would recommend trying out t3.chat ( that’s one of the platforms that I use). Cost 8 USD / month and is made by Theo pretty happy with it for the price. The UI is honestly its strongest point.

For how I actually use AI, I wrote a more detailed answer in another thread about AI usage. Have a read

A bit of a tangent but I wonder if this is becoming more common these days. The “shut-in” phenomena, japanese call them hikikomori. News headlines say it is but i wonder how well they represent reality.

There is a fifth way of using AI: Ask AI to hint the problems of your text or suggest rules to look up so you can can “solve” it yourself.

The problem of over-reliance on AI isn’t anything new - we’ve always learned by struggling through problems ourselves. It’s like playing a puzzle game - if you just go look up the solutions instead of trying things yourself, not only do you lose the point of playing the game by reducing it to a series of bothersome tasks that just need to get done to get something at the end, but eventually you find yourself out of depth as you didn’t develop the proper understanding of the puzzles.

Because of this, I’ve been gravitating more towards the 4th and 5th ways of using AI for things that matter to me, things I need or want to understand deeply.

I try not to rush through things, to enjoy the process and instead of just asking for an answer to a question, I’m starting to ask it: ** How can I find the answer myself? What materials would an experienced person in this field look up in order to solve this problem?** And similar variations of these types of questions. The main idea is: I instruct it not to give a solution or code right away but instead to explore the problem together with me and teach me how to fish instead of giving me the fish. If I give him some part of the documentation and he gets an insight, I ask: How did this part of the docs help you get to the conclusion of that? How did you know what to look for ? And so on. Basically I assume the AI is a more experienced person next to me and we’re trying to pair program. He doesn’t know the solution from the back of his mind but he can easily “find it” and we’re walking through it together.

This shift happened because AI kept missing the mark on my questions - partly because I work with relatively niche tools, partly because when you’re learning, you don’t know what context is even relevant to give it and if you give irrelevant context usually you end up misleading it.

And it’s actually surprisingly fun and enjoyable to work on my problems now. There’s this shift of not seeing the problems as something to be solved but as something that needs to be understood, a game that needs to be played if you will. Obviously it takes longer as the article pointed out learning takes time.

It’s been awhile since I had to deal with web requests. Back then I used Postman (back when I still used Windows). Now I try to always get by using the simplest open source tools for the job so thanks for sharing, will try it out.

I knew of curl but always thought of it as a tool to play around when doing simple requests, never knew people go so far with it.